• Aucun résultat trouvé

O VI in the Local Interstellar Medium

N/A
N/A
Protected

Academic year: 2021

Partager "O VI in the Local Interstellar Medium"

Copied!
26
0
0

Texte intégral

(1)

HAL Id: hal-00531351

https://hal.archives-ouvertes.fr/hal-00531351

Submitted on 10 Jul 2020

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

O VI in the Local Interstellar Medium

M. A. Barstow, D. D. Boyce, J. K. Barstow, A. E. Forbes, B. Y. Welsh,

Rosine Lallement, S. Preval

To cite this version:

M. A. Barstow, D. D. Boyce, J. K. Barstow, A. E. Forbes, B. Y. Welsh, et al.. O VI in the

Lo-cal Interstellar Medium. The AstrophysiLo-cal Journal, American AstronomiLo-cal Society, 2010, 723 (2),

pp.1762-1786. �10.1088/0004-637X/723/2/1762�. �hal-00531351�

(2)

C

2010. The American Astronomical Society. All rights reserved. Printed in the U.S.A.

O vi IN THE LOCAL INTERSTELLAR MEDIUM

M. A. Barstow1, D. D. Boyce1, B. Y. Welsh2, R. Lallement3, J. K. Barstow1,4, A. E. Forbes1, and S. Preval1 1Department of Physics and Astronomy, University of Leicester, University Road, Leicester LE1 7RH, UK

2Experimental Astrophysics Group, Space Sciences Laboratory, University of California Berkeley, Berkeley, CA 94720, USA 3Service d’A´eronomie du CNRS, 91371 Verri`eres-le-Buisson, France

4Atmospheric, Oceanic and Planetary Physics, Department of Physics, Clarendon Laboratory, Parks Road, Oxford OX1 3PU, UK

Received 2009 March 4; accepted 2010 September 13; published 2010 October 26

ABSTRACT

We report the results of a search for O vi absorption in the spectra of 80 hot DA white dwarfs observed by the FUSE satellite. We have carried out a detailed analysis of the radial velocities of interstellar and (where present) stellar absorption lines for the entire sample of stars. In approximately 35% of cases (where photospheric material is detected), the velocity differences between the interstellar and photospheric components were beneath the resolution of the FUSE spectrographs. Therefore, in 65% of these stars the interstellar and photospheric contributions could be separated and the nature of the O vi component unambiguously determined. Furthermore, in other examples, where the spectra were of a high signal-to-noise, no photospheric material was found and any O vi detected was assumed to be interstellar. Building on the earlier work of Oegerle et al. and Savage & Lehner, we have increased the number of detections of interstellar O vi and, for the first time, compared their locations with both the soft X-ray background emission and new detailed maps of the distribution of neutral gas within the local interstellar medium. We find no strong evidence to support a spatial correlation between O vi and SXRB emission. In all but a few cases, the interstellar O vi was located at or beyond the boundaries of the local cavity. Hence, any T∼ 300,000 K gas responsible for the O vi absorption may reside at the interface between the cavity and surrounding medium or in that medium itself. Consequently, it appears that there is much less O vi-bearing gas than previously stated within the inner rarefied regions of the local interstellar cavity.

Key words: ISM: clouds – ISM: general – ISM: structure – ultraviolet: ISM – ultraviolet: stars – white dwarfs Online-only material: color figures

1. INTRODUCTION

The existence of the diffuse soft X-ray background (SXRB; Bowyer et al. 1968) implies that a substantial fraction of the Galactic disk, at least near the Sun, is filled with low-density hot gas (McCammon et al.1983). This discovery has important implications for our understanding of the structure and ioniza-tion of the interstellar medium (ISM) and in subsequent decades further studies have been made of the diffuse background, with a full sky survey being conducted by the ROSAT mission (Snowden et al.1995,1998). The total soft X-ray emission has been interpreted as the sum of three distinct components; extra-galactic, galactic halo, and Local Bubble (LB) hot gas contribu-tions. Details of these models and the constraints on the location of the soft X-ray emitting gas deduced from a number of soft X-ray shadows are discussed by Snowden et al. (1998,2000) and Kuntz & Snowden (2000), indicating that the temperature of the local hot gas is T≈ 106K. If this gas is homogeneously distributed in the LB, the measured emission in any direction must be proportional to the path length through the hot gas to the surrounding dense, cool material. Therefore, it is possible to map the shape of the emitting region (Snowden et al.1998). Among the main features are two large elongations toward op-posite directions at high galactic latitudes. Their axis is tilted by about 20◦–30◦with respect to the galactic polar axis, coinciding with neutral gas-free “chimneys” found nearly perpendicular to the Gould belt (Sfeir et al.1999; Welsh et al.1999).

Complementary to the study of the diffuse SXRB are obser-vations of absorption from the lithium-like O vi ion transition doublet in the far-ultraviolet at 1031.91 and 1037.62 Å, arising in the ISM (Rogerson et al.1973; York1974; Jenkins & Meloy

1974; Jenkins1978). Ionization of O v to O vi requires an

en-ergy input of 114 eV. Since the local interstellar radiation field declines steeply at energies above this value (Vallerga1998), it is very difficult to account for the presence of O vi with a pho-toionization model. Therefore, the best explanation is that the O vi arises from collisional ionization in gas at a temperature in excess of 2× 105 K. This proposition is supported by the relative breadth of the O vi absorption line profiles, compared to those of lower ionization potential, which are consistent with thermal broadening at this temperature. In this paper, we refer to the T ∼ 300,000 K material which may be responsible for O vi absorption as “transition temperature” gas to differentiate it from the “hot”∼1,000,000 K X-ray emitting gas.

The evidence for the presence of hot gas in the LB seems to be strong. However, there are a number of other observational results that contradict this picture. Any hot gas in the local ISM should give rise to a diffuse emission which should be detected not just in the soft X-ray waveband, but also in the extreme ultraviolet. Analysis of the ROSAT Wide Field Camera (WFC) EUV sky survey data reveals that the diffuse EUV background is negligible compared to other sources (West et al. 1994). Similarly, attempts to detect EUV line emission from the hot gas in deep exposures with the Extreme-Ultraviolet Explorer (EUVE) spectrographs yielded only upper limits to the flux (Vallerga & Slavin 1998). Both the ROSAT WFC and EUVE telescopes were optimized for the detection and observation of point sources rather than diffuse emission. In contrast, the

Cosmic Hot Interstellar Plasma Spectrometer was specifically

designed to detect the line emission from the putative hot interstellar gas component. Again, no background emission was detected, although very tight constraints were put on the possible plasma emission measure (Hurwitz et al.2005) placing it an order of magnitude lower than implied by the soft X-ray 1762

(3)

fluxes. Either there is less hot gas than expected, and the SXRB has another source, or the hot gas is significantly depleted in iron relative to other heavy elements, much more than reported elsewhere for hot gas.

One of the surprise results from the ROSAT mission was the detection of X-ray emission from the comet Hayakutake (Lisse et al. 1996), subsequently interpreted as arising from charge transfer of heavy solar wind ions with cometary neutrals (Cravens 1997). This solar wind charge exchange (SWCX) mechanism should also be experienced by neutral atoms of interstellar origin within the solar system. De-excitation of charge-exchanged atoms should then generate diffuse soft X-ray emission throughout the heliosphere (Cox1998; Cravens

2000). Long-term enhancements in the background observed by ROSAT appear to be correlated with solar activity (Freyberg

1994, Cravens2000), indicating that a fraction of the SXRB arises from the heliosphere rather than the LB. Lallement (2004) finds that the fraction of SXRB emission attributable to our local cavity is substantially lower than previously thought.

Nevertheless, recent measurements of O vi absorption in the spectra of hot white dwarfs observed by the Far-Ultraviolet

Spectroscopic Explorer (FUSE) mission provide renewed

evi-dence for the presence of significant quantities of hot gas in the LB (Oegerle et al.2005; Savage & Lehner2006). These authors report positive detections of O vi absorption along several lo-cal lines of sight but, as pointed out in these papers, there is the problem of possible contamination from photospheric ma-terial in the hottest white dwarfs making interpretation of the spectral measurements difficult. Indeed, International

Ultravi-olet Explorer (IUE) and Hubble Space Telescope (HST)

high-resolution UV observations of DA white dwarfs show that high ionization species such as N v are found in the photospheres of white dwarfs down to surprisingly low effective temperatures. Originally, resonance lines of N v, C iv, and Si iv detected in the IUE echelle spectrum of GD659 (Teff= 35,500 K), reported by Holberg et al. (1995) were thought to be circumstellar in origin. However, a new measurement of the stellar radial veloc-ity shows that these features actually arise in the photosphere (Holberg et al.2000). The same resonance lines are also de-tected at the photospheric velocity of REJ1032+532 (Holberg et al.1999; Teff= 46,330 K) and photospheric N v is seen in the HST/GHRS spectrum of REJ1614−085 (Holberg et al.1997;

Teff= 38,500 K), indicating that it is not possible to assume that

high ionization species are not photospheric in the cooler (i.e.,

Teff< 50,000 K) hot DA stars.

An additional concern is the apparent absence of detections of other high-ionization interstellar species in the UV spectra of white dwarfs. At gas temperatures sufficiently high to populate O vi, we might also expect to see N v and C iv. The ratio of the populations of N v and C iv to O vi will depend on the gas temperature, but for typical interstellar structures this lies somewhere in the range 0.1–0.4 (Indebetouw & Shull2004), indicating that the N v and C iv resonance doublets should be present if O vi is detected. However, if the O vi lines are weak, the N v and C iv may fall below their detection threshold. From an observational point of view, there are no detections of interstellar N v or C iv along the lines of sight to any local white dwarf observed with IUE (Holberg et al.1998). Furthermore, careful radial velocity analyses of a smaller but higher signal-to-noise sample of DA white dwarfs studied with HST/STIS showed detections of both photospheric and circumstellar N v and C iv (Barstow et al.2003b; Bannister et al.2003), but no interstellar features.

Savage & Lehner (2006) take the reasonable position that if no stellar or circumstellar heavy element lines are detected in a spectrum then any detected O vi is interstellar in origin. Ruling out a photospheric component is straightforward, since there are lines of C iii/iv, O iv, Si iv, P iv/v, and S iv/VI that are readily detectable in the FUSE band. It is harder to be sure that there is no circumstellar component because there are no important resonance lines in the FUSE band other than the O vi. Fortunately, there are existing data from IUE and HST for a number of stars that can be used to confirm the absence of circumstellar material (e.g., WD1314+293, WD1254+223) and the work of Bannister et al. (2003) indicates that circumstellar components are only present when there is also photospheric material in a given star.

When stellar metals are present, following Oegerle et al. (2005) and Savage & Lehner (2006), the only way to unam-biguously determine the nature of a detection of O vi in FUSE white dwarf spectra is to carry out a thorough analysis of the ra-dial velocity of known interstellar and photospheric features for each individual star for comparison with the O vi velocity. Even then the difference between the interstellar and photospheric velocities must be greater or equal to the velocity resolution of the instrument. Few stars which have photospheric features have sufficient velocity separation between these and the ISM lines. Hence, even samples of several tens of objects, as studied by Oegerle et al. (2005) and Savage & Lehner (2006) are whittled down to just a few good targets. Since the work reported in these papers was carried out, more observations have been added to the FUSE archive, increasing the number of stars available for study. Often multiple observations of a single target are available and can be co-added to improve the resultant signal-to-noise ra-tio (S/N). Furthermore, the archival data are now available with an improved version of the FUSE pipeline (version 3.2, com-pared to version 2.05 and version 1.8.7 used by Savage & Lehner and Oegerle et al., respectively), which (hopefully) yields better quality data in terms of S/N and calibration.

It is therefore timely to re-examine the incidence of O vi in hot white dwarfs observed by FUSE and where present investigate its nature. We report here on such a new study including a sample of 80 stars drawn from the FUSE archive. Of this group, we have detected O vi in 26 cases where there is either no confusing photospheric component or when the interstellar and photospheric velocities are well separated. When we compare the distribution of stars in this study with neutral gas density maps of the local ISM it appears that the majority of interstellar O vi detections appear to be located at the periphery of or outside the LB. Hence, there is little evidence for the presence of million ◦K gas within the LB. In this paper, we describe our analysis of the white dwarf sample in detail and consider the implications our results have for our understanding of the local ISM and the origin of the SXRB.

2. THE WHITE DWARF SAMPLE

We have surveyed all the hot DA white dwarfs (Teff > 20,000 K) observed by FUSE to look for evidence of the pres-ence of interstellar O vi absorption. The total sample comprises 95 stars and includes all public observations accessible in the

FUSE data archive up to 2007 August 1. However, while there

are many observations of cooler DAs and also a substantial number of observations of He-rich DO white dwarfs, we have excluded these targets from our study for various reasons. In the DA stars, at temperatures below ∼20,000 K, the far-UV continuum flux is severely attenuated by the Lyman line series

(4)

absorption in the spectral range occupied by the O vi doublet which yields an S/N below acceptable limits for all reasonable exposures. In some stars, the continuum flux is contaminated by the presence of absorption by interstellar H2, leading to confu-sion in identifying the lines we need to measure for this work or reducing the S/N of the data below useful levels. In princi-ple, the DO white dwarfs, which all have temperatures above ∼45,000 K, constitute excellent targets for this work. However, there are often many high-ionization features present in these for which the atomic data are not very reliable making it hard to determine reliable radial velocities (see Barstow et al.2007), in contrast to the DAs. In addition, these objects potentially contain significant abundances of Fe group elements with fea-tures distributed throughout the FUSE wavelength range which might significantly contaminate or mimic a weak detection of O vi. Our resulting target sample of 80 stars has considerable overlap with both those of Oegerle et al. (2005) and Savage & Lehner (2006), but is factors 3.6 and 1.7 larger, respectively. It is important to note that, because of the selection criteria we have outlined above, we have not included all the stars examined by these authors in this paper. On the other hand, they excluded from their analyses previously known stars where the O vi is thought to be contaminated or where the S/N is not adequate, which we do consider here.

All the white dwarfs we have considered in our detailed anal-ysis are listed in Table1, along with their physical parameters (Teff, log g, mass, and distance), where known, and their galac-tic coordinates. The targets include both isolated white dwarfs and also binary systems where the white dwarf has a main-sequence companion (mostly those objects with an HD number). For several binaries, no published white dwarf temperatures and gravities exist, so we have estimated these values from the Lyman lines present in the FUSE spectra, following the methods outlined in Barstow et al. (2003a). We have also updated some values where original estimates were made using just the Lyα line in IUE SWP spectra. The list is ordered by the white dwarf R.A/decl. number, where one has been assigned in the McCook & Sion (1999) catalog, and any principal alternative name in general use is also given. Where there is no such name, the designation of the primary discovery survey is noted instead.

3. OBSERVATIONS AND DATA REDUCTION

3.1. The FUSE Spectroscopic Data

The FUSE mission was launched in 1999 and was success-fully operated on-orbit until 2007. The instrument design and in-flight performance have been described in several scien-tific papers, including Moos et al. (2000) and Sahnow et al. (2000). The most recent version of the data processing pipeline, CALFUSE version 3, is described by Dixon et al (2007). Briefly, the spectrographs are based on a Rowland circle design, com-prising four separate optical paths (or channels). The channels must be co-aligned so that light from a single target properly illuminates all four channels to maximize the throughput of the instrument. Spectra from the four channels are recorded on two microchannel plate detectors, with an SiC and LiF spectrum on each. Each detector is divided into two functionally independent segments (A and B), separated by a small gap. Consequently, there are eight detector segment/spectrometer channel combi-nations to be dealt with in reducing the data. Maintaining the co-alignment of individual channels has been difficult in-orbit, mainly due to thermal effects. A target may completely miss an aperture for the whole or part of an observation, while being

well centered in the others. Hence, in any given observation, not all of the channels may be available in the data. To minimize this problem, most observations have been conducted using the largest aperture available (LWRS, 30× 30 arcsec). This lim-its the spectral resolution to between 15,000 and 20,000 for early observations or 23,000 later in the program when the mirror focusing had been adjusted (Sahnow et al.2000), com-pared to the 24,000–30,000 expected for the 1.25× 20 arcsec HIRS aperture. Various spectra analyzed here were obtained through HIRS, MDRS, or LWRS apertures and in TIMETAG or HISTOGRAM mode. Table2summarizes the data sets used for each of the white dwarf spectra studied in this paper.

3.2. Data Reduction

All the data used in this work were obtained from the archive (see Table2) after reprocessing using version 3.2 of the CALFUSEpipeline. The data supplied by the archive consist of a set of files corresponding to the separate exposures for each of the eight channel/detector segment combinations. Since the S/N of these can be relatively poor and the wavelength binning (∼0.006 Å) over-samples the true resolution by a factor ∼10, we have established a tried and tested procedure to combine all the spectra (see Barstow et al.2003a). Initially all the spectra were re-binned to a 0.02 Å pixel size for examination. Any strong interstellar absorption features present are used to verify that the wavelength scales for each exposure are well aligned. Individual exposures are then co-added to produce a single spectrum, weighting the individual spectra by their exposure time. This whole procedure was repeated for all eight detector/ channel side combinations.

In principle, since we are mainly concerned with the analysis of the O vi lines in this paper, it is not necessary to study any segments other than the LiF 1A, SiC 1A, LiF 2B, and SiC 2B, which encompass the O vi doublet. However, for this work we also need to make accurate measurements of the interstellar and photospheric velocities. Ideally, these would simply be obtained from photospheric and interstellar features immediately adjacent the O vi lines. Unfortunately, in many of the stars in our sample this is not practical, since there are no suitable lines. There may be no photospheric lines at all in the segment, or the interstellar lines are saturated and/ or contaminated with geocoronal emission preventing accurate determination of the line velocity.

The perceived radial velocity of the photospheric lines will be the combination of both the intrinsic radial velocity of the star and its gravitational redshift, the latter depending on the stellar mass and radius. This could be measured at other wavelengths for our program stars, but few reliable measurements are available for them in the literature. Both the IUE echelle and HST/STIS spectrographs had extremely good wavelength calibration and provide the best data but only ∼25% of our sample have been observed by these instruments (see Barstow et al.2003b; Bannister et al.2003). Even in these stars, assigning a photospheric velocity to the FUSE data is a tricky issue because the FUSE wavelength scale zero point is not necessarily the same for each exposure and spectral segment due to flexure in the instrument and the general use of the largest LWRS aperture to compensate, which leads to a significant uncertainty in the positioning of a source within it. As a result, the accuracy of the

FUSE wavelength scale is typically limited to ∼15 km s−1. However, comparisons between the IUE echelle and FUSE for GD71 indicate that this uncertainty could be as large as ∼20 km s−1(Dobbie et al.2005). Therefore, we must be careful

(5)

Table 1

Summary of White Dwarfs Targets and their Physical Parameters

WD No. Alt. Name/Cata L (◦) b (◦) d (pc) Reference Teff log g Reference

WD0001+433 REJ, EUVEJ 113.90 −18.44 99 This work 46205 8.85 Marsh et al.1997 WD0004+330 GD2 112.48 −28.69 99 Vennes et al.1997 47936 7.77 Marsh et al.1997 WD0027−636 REJ, EUVEJ 306.98 −53.55 236 Vennes et al.1997 60595 7.97 Marsh et al.1997 WD0050−332 GD659 299.14 −84.12 58 Holberg et al.1998 34684 7.89 Marsh et al.1997 WD0106−358 GD683 280.88 −80.81 86 This work 28580 7.90 Marsh et al.1997 WD0131−164 GD984 167.26 −75.15 96 Vennes et al.1997 44850 7.96 Marsh et al.1997 WD0147+674 REJ, EUVEJ 128.58 5.44 99 Dupuis et al.2005 30120 7.70 Marsh et al.1997 WD0226−615 HD15638 284.20 −52.16 199 ESA1997 50000 8.15 Landsman et al.1993 WD0229−481 REJ, EUVEJ 266.62 −61.59 240 This work 63400 7.43 Marsh et al.1997 WD0232+035 Feige 24 165.97 −50.27 74 ESA1997 62947 7.53 Marsh et al.1997 WD0236+498 REJ, EUVEJ 140.15 −9.15 96 Vennes et al.1997 33822 8.47 Finley et al.1997 WD0235−125 PHL1400 165.97 −50.27 86 This work 32306 8.44 Finley et al.1997 WD0252−055 HD18131 181.86 −53.47 104 ESA1997 29120 7.50 Vennes et al.1997 WD0320−539 LB1663 267.30 −51.64 124 This work 32860 7.66 Marsh et al.1997 WD0325−857 REJ, EUVEJ 299.86 −30.68 35 Barstow et al.1995 . . . . . . . . .

WD0346−011 GD50 188.95 −40.10 29 Vennes et al.1997 42373 9.00 Marsh et al.1997 WD0353+284 REJ, EUVEJ 165.08 −18.67 107 Burleigh et al.1997 32984 7.87 This work WD0354−368 238.64 −49.98 400 Christian et al.1996 53000 8.00 Christian et al.1996 WD0455−282 REJ, EUVEJ 229.29 −36.17 102 Oegerle et al.2005 58080 7.90 Marsh et al.1997 WD0501+524 G191-B2B 155.95 7.10 59 Vennes et al.1997 57340 7.48 Marsh et al.1997 WD0512+326 HD33959C 173.30 −3.36 25 ESA1997 22750 8.01 Holberg et al.1999 WD0549+158 GD71 192.03 −5.34 49 Vennes et al.1997 32780 7.83 Barstow et al.2003a WD0603−483 REJ, EUVEJ 255.78 −27.36 178 Vennes et al.1997 33040 7.80 Marsh et al.1997 WD0621−376 REJ, EUVEJ 245.41 −21.43 78 Holberg et al.1998 62280 7.22 Marsh et al.1997 WD0659+130 REJ0702+129 202.51 8.19 115 Vennes et al.1998 39960 8.31 This work WD0715−704 REJ, EUVEJ 281.40 −23.50 94 Vennes et al.1997 44300 7.69 Marsh et al.1997 WD0802+413 KUV 179.22 30.94 230 Dupuis et al.2005 45394 7.39 Finley et al.1997 WD0830−535 REJ, EUVEJ 270.11 −8.27 82 Vennes et al.1997 29330 7.79 Marsh et al.1997 WD0937+505 PG 166.90 47.12 218 Dupuis et al.2005 35552 7.76 Finley et al.1997 WD1019−141 REJ, EUVEJ 256.48 34.74 112 This work 29330 7.79 Marsh et al.1997 WD1021+266 HD90052 205.72 57.21 250 Burleigh et al.1997 35432 7.48 This work WD1024+326 REJ, EUVEJ 194.52 58.41 400 This work 41354 7.59 This work WD1029+537 REJ, EUVEJ 157.51 53.24 116 Vennes et al.1997 44980 7.68 Marsh et al.1997 WD1040+492 REJ, EUVEJ 162.67 57.01 230 Vennes et al.1997 47560 7.62 Marsh et al.1997 WD1041+580 REJ, EUVEJ 150.12 52.17 93 Vennes et al.1997 29016 7.79 Marsh et al.1997 WD1057+719 PG 134.48 42.92 141 Savage & Lehner2006 39555 7.66 Marsh et al.1997 WD1109−225 HD97277 274.78 34.54 82 ESA1997 36750 7.50 Barstow et al.1994 WD1234+481 PG 129.81 69.01 129 Vennes et al.1997 55570 7.57 Marsh et al.1997 WD1254+223 GD153 317.25 84.75 67 Vennes et al.1997 38205 7.90 Barstow et al.2003a WD1302+597 GD323 119.82 57.59 79 This work 39960 8.31 This work WD1314+293 HZ43 54.11 84.16 68 Finley et al.1997 49435 7.95 Barstow et al.2003a WD1337+701 EG102 117.18 46.31 104 Dupuis et al.2005 20435 7.87 Finley et al.1997 WD1342+442 PG 94.30 69.92 387 This work 66750 7.93 Barstow et al.2003a WD1440+753 REJ, EUVEJ 114.10 40.12 98 Vennes et al.1997 42400 8.54 Vennes et al.1997 WD1528+487 REJ, EUVEJ 78.87 52.72 140 Vennes et al.1997 46230 7.70 Marsh et al.1997 WD1550+130 NN Ser 23.65 45.34 756 This work 39910 6.82 This work WD1603+432 PG 68.22 47.93 114 Dupuis et al.2005 36257 7.85 Finley et al.1997 WD1611−084 REJ, EUVEJ 4.30 29.30 93 Vennes1999 38500 7.85 Marsh et al.1997 WD1615−154 G153-41 358.79 24.18 55 Oegerle et al.2005 38205 7.90 Barstow et al.2003a WD1620+647 PG 96.61 40.16 174 Dupuis et al.2005 30184 7.72 Finley et al.1997 WD1631+781 REJ, EUVEJ 111.30 33.58 67 Vennes et al.1997 44559 7.79 Finley et al.1997 WD1635+529 HD150100 80.94 41.49 123 ESA1997 20027 8.14 This work WD1636+351 WD1638+349 56.98 41.40 109 Vennes et al.1997 36056 7.71 Marsh et al.1997 WD1648+407 REJ, EUVEJ 64.64 39.60 200 Vennes et al.1997 37850 7.95 Marsh et al.1997 WD1725+586 LB335 87.17 33.83 123 This work 54550 8.49 Marsh et al.1997 WD1800+685 KUV 98.73 29.78 159 Vennes et al.1997 43701 7.80 Marsh et al.1997 WD1819+580 REJ, EUVEJ 87.00 26.76 103 This work 45330 7.73 Marsh et al.1997 WD1844−223 REJ, EUVEJ 12.50 −9.25 62 Vennes et al.1997 31470 8.17 Finley et al.1997 WD1845+683 KUV 98.81 25.65 125 Vennes et al.1997 36888 8.12 Finley et al.1997 WD1917+509 REJ, EUVEJ 91.02 20.04 105 Vennes et al.1997 33000 7.90 Vennes et al.1997 WD1921−566 REJ, EUVEJ 340.63 −27.01 110 Vennes et al.1998 52946 8.16 This work WD1942+499 83.08 12.75 105 Vennes et al.1997 33500 7.86 Marsh et al.1997 WD1950−432 HS 356.51 −29.09 140 Dupuis et al.2005 41339 7.85 Finley et al.1997 WD2000−561 REJ, EUVEJ 341.78 −32.25 198 Dupuis et al.2005 44456 7.54 Marsh et al.1997 WD2004−605 REJ, EUVEJ 336.58 −32.86 58 Vennes et al.1997 44200 8.14 Marsh et al.1997 WD2011+398 REJ, EUVEJ 77.00 3.18 141 Vennes et al.1997 47057 7.74 Marsh et al.1997

(6)

Table 1

(Continued)

WD No. Alt. Name/Cata L () b () d (pc) Reference Teff log g Reference

WD2014−575 LS210-114 340.20 −34.25 51 Vennes et al.1997 26579 7.78 Marsh et al.1997 WD2020−425 REJ, EUVEJ 358.36 −34.45 52 This work 28597 8.54 Marsh et al.1997 WD2111+498 GD394 91.37 1.13 50 Vennes et al.1997 38866 7.84 Marsh et al.1997 WD2116+736 KUV 109.39 16.92 177 Dupuis et al.2005 54486 7.76 Finley et al.1997 WD2124+191 HD204188 70.43 −21.98 46 Vennes et al.1998 35000 9.00 Landsman et al.1993 WD2124−224 REJ, EUVEJ 27.36 −43.76 224 Oegerle et al.2005 48297 7.69 Marsh et al.1997 WD2146−433 REJ, EUVEJ 357.18 −50.13 362 Dupuis et al.2005 67912 7.58 Finley et al.1997 WD2152−548 REJ, EUVEJ 340.50 −48.70 129 . . . 45800 7.78 Vennes et al.1997 WD2211−495 REJ, EUVEJ 345.79 −52.62 53 Vennes et al.1997 65600 7.42 Marsh et al.1997 WD2257−073 HD217411 65.17 −56.93 89 Vennes et al.1998 38010 7.84 This work WD2309+105 GD246 87.26 −45.12 79 Vennes et al.1997 51300 7.91 Barstow et al.2003a WD2321−549 REJ, EUVEJ 326.91 −58.21 192 Dupuis et al.2005 45860 7.73 Marsh et al.1997 WD2331−475 REJ, EUVEJ 334.84 −64.81 82 Vennes et al.1997 56682 7.64 Marsh et al.1997 WD2350−706 HD223816 309.91 −45.94 92 Barstow et al.2001 69300 8.00 Barstow et al.1994

Note.aPG, Palomar Green; HS: Hamburg Schmidt; REJ, ROSAT EUV; EUVEJ, EUVE; KUV, Kiso UV; KPD, Kitt Peak Downes.

when directly comparing a velocity reference from IUE or HST with a FUSE measurement of an O vi velocity.

For the purposes of this analysis we do not actually need a very accurate absolute measurement of the radial velocities. What is most important is that we are able to obtain an accurate measurement of the difference between the interstellar and stellar components and an accurate determination of the O vi velocity on the same scale. Any undetermined offset to the wavelength zero point would be unimportant provided it is the same for all data segments. Therefore, to be able to obtain good relative velocities it was necessary to combine the individual spectral segments to provide continuous coverage across the full FUSE spectrum. First, all the spectra were re-sampled onto a common wavelength scale of 0.02 Å steps to reduce the level of oversampling. The re-sampled/re-binned spectra were then co-added, weighting individual data points by the statistical variance, averaged over a 20 Å interval, to take into account any differences in the effective area of each segment and any differences in the exposure time that may have arisen from rejection of bad data segments (e.g., those with reduced exposure due to source drift). We have used the measured wavelengths of the strongest unsaturated interstellar lines to adjust the wavelength scale for each segment before co-addition, taking the LiF 1A as our reference point against which every other segment is adjusted.

Through visual inspection, it is apparent that the statistical noise tends to increase toward the edges of a wavelength range. In cases where the S/N is particularly poor (<3:1 per resolution element) in these regions, we have trimmed the spectra (by ∼2–3 Å) to remove these data points prior to co-addition. An example of the resulting spectra, taking one of the higher S/N data sets, is shown in Figure1. The region of poor S/N seen in the 1080–1087 Å region is due to the absence of LiF data at those wavelengths. Where we had multiple exposures for any white dwarf, we carried out a further stage of evaluation of the spectra, discarding those where the S/N was particularly poor or where the flux levels appeared to be anomalous, when compared to the remainder.

3.3. Measuring Photospheric and Interstellar Radial Velocities

The output of our data reduction procedure is a composite, wavelength calibrated spectrum covering the full range of the FUSE spectrographs for each white dwarf in the sample. The data for individual stars can then be visually searched

for absorption features across the full wavelength range. In a separate exercise, we have cross-correlated the measured wavelengths of detected features with all the publicly available line lists (e.g., Kurucz & Bell 1995; Morton 2003; Kentucky database—http://www.pa.uky.edu/∼peter/atomic). A particular problem for line identification in the far-UV is the large number of possible transitions for feasible species, often leading to several alternative nearby identifications (i.e., within the measurement and wavelength scale uncertainties) for a single feature. By requiring that all reliable identifications correspond to one of two possible radial velocities, either interstellar or photospheric, in an individual star and searching for similar absorption patterns in a number of stars, we have produced a master list for line identification purposes (Boyce2009). This forms the basis for the determination of interstellar and stellar radial velocities in this paper.

For each star in the sample, we have searched for signifi-cant features across the full FUSE range and measured their wavelengths using a simple determination of the centroid of the feature by fitting a Gaussian profile. In general, it is not necessary or appropriate to use more sophisticated line fitting procedures due to the undersampling of the lines by the FUSE spectral resolution and the relatively low S/N of most features. The main purpose of this paper is to compare the observed velocity of any O vi features detected in these white dwarf spec-tra with the interstellar and stellar velocities to determine with which particular component, if any, it is associated. Ideally, to avoid any issues associated with differing wavelength cali-brations for different spectroscopic segments, the comparison should be against lines that are close to O vi in wavelength and in the same spectral segment. The available interstellar ref-erence lines are C ii (1036.337 Å), O i (1039.230 Å), and Ar i (1048.220 Å), but there are few potential photospheric lines in the region apart from the O vi itself. Several P iv lines (1030.515, 1033.11, and 1035.516) fall in this wavelength range, but are present in only a few stars. Furthermore, any O i absorption may be contaminated by geocoronal emission and the C ii can be saturated, compromising the wavelength measurement in ei-ther case. If we were to rely on velocity comparisons in the O vi region alone, we would have very few stars available for this study. Therefore, we have determined the mean interstellar and stellar velocities for each star from all the relevant lines available across the full FUSE spectral range. These results are summarized in Figures2and3, and Table3, together with the

(7)

Table 2

Summary of FUSE Observations

WD No. Alt. Name/Cat Observation Start Time (yyyy-mm-dd) WD0001+433 REJ, EUVEJ E90305010 2004-11-24 19:34:00 WD0004+330 GD2 P20411020-40 2002-09-14 10:07:00 WD0027−636 REJ, EUVEJ Z90302010 2002-09-30 15:44:00 WD0050−332 GD659 M10101010 2000-07-04 23:52:00 P20420010 2000-12-11 07:00:00 WD0106−358 GD683 D02301010 2004-11-21 06:29:00 WD0111+002 A01303030 2001-08-07 12:43:00 WD0131−164 GD984 P20412010 2000-12-10 07:45:00 WD0147+674 REJ, EUVEJ Z90303020 2002-10-12 17:08:00 WD0226−615 HD15638 A05402010 2000-07-04 20:39:00 WD0229−481 REJ, EUVEJ M10504010 2002-09-21 11:51:00 WD0232+035 Feige 24 P10405040 2004-01-06 01:00:00 WD0236+498 REJ, EUVEJ Z90306010 2002-12-11 16:47:00 WD0235−125 PHL1400 E56809010 2004-08-30 17:27:00 WD0252−055 HD18131 A05404040 2001-11-28 06:39:00 WD0320−539 LB1663 D02302010 2003-09-09 08:50:00 WD0325−857 REJ, EUVEJ A01001010 2000-09-17 19:52:00 WD0346−011 GD50 B12201020-50 2003-12-20 04:43:00 WD0353+284 REJ, EUVEJ B05510010 2001-01-02 22:31:00 WD0354−368 B05511010 2001-08-12 11:18:00 WD0416+402 KPD Z90308010 2003-10-05 13:39:00 WD0455−282 REJ, EUVEJ P10411030 2000-02-07 09:36:00 WD0501+524 G191-B2B Coadded∗ WD0512+326 HD33959C A05407070 2000-03-01 04:08:00 WD0549+158 GD71 P20417010 2000-11-04 15:32:00 WD0603−483 REJ, EUVEJ Z90309010 2002-12-31 19:26:00 WD0621−376 REJ, EUVEJ P10415010 2000-12-06 05:28:00 WD0659+130 REJ0702+129 B05509010 2001-03-05 21:53:00 WD0715−704 REJ, EUVEJ M10507010 2003-08-15 00:08:00 WD0802+413 KUV Z90311010 2004-03-14 09:37:00 WD0830−535 REJ, EUVEJ Z90313010 2002-05-03 13:37:00 WD0937+505 PG Z90314010 2003-03-24 09:13:00 WD1019−141 REJ, EUVEJ P20415010 2001-05-15 21:35:00 WD1021+266 HD90052 B05508010 2001-05-01 19:34:00 WD1024+326 REJ, EUVEJ B05512010 2001-05-01 09:22:00 WD1029+537 REJ, EUVEJ B00301010 2001-03-25 04:40:00 WD1040+492 REJ, EUVEJ Z00401010 2002-04-04 12:42:00 WD1041+580 REJ, EUVEJ Z90317010 2002-04-07 10:04:00 WD1057+719 PG Z90318010 2002-04-08 16:12:00 WD1109−225 HD97277 A05401010 2000-05-29 16:00:00 WD1234+481 PG M10524020 2003-03-18 20:29:00 WD1254+223 GD153 M10104020 2000-04-29 21:32:00 WD1302+597 GD323 C02601010 2002-04-02 17:00:00 WD1314+293 HZ43 P10423010 2000-04-22 21:19:00 WD1337+701 EG102 B11903010 2001-05-05 06:43:00 WD1342+442 PG A03404020 2000-01-11 23:01:00 WD1440+753 REJ, EUVEJ Z90322010 2003-01-15 08:40:00 WD1528+487 REJ, EUVEJ P20401010 2001-03-27 23:14:00 WD1550+130 NN Ser E11201010 2004-04-15 05:38:00 WD1603+432 PG Z90324010 2003-07-02 11:51:00 WD1611−084 REJ, EUVEJ B11904010 2004-04-02 11:53:00 WD1615−154 G153-41 P20419010 2001-08-29 12:40:00 WD1620+647 PG Z90325010 2002-05-17 04:45:00 WD1631+781 REJ, EUVEJ M10528040 2004-02-13 20:02:00 WD1635+529 HD150100 C05002010 2002-07-13 14:00:00 WD1636+351 WD1638+349 P20402010 2001-03-28 14:22:00 WD1648+407 REJ, EUVEJ Z90326010 2002-07-12 11:22:00 WD1725+586 LB335 Z90327010 2002-05-13 23:16:00 WD1800+685 KUV M10530010-70 2002-09-09 15:14:53 P20410010-20 2001-10-01 04:55:36 WD1819+580 REJ, EUVEJ Z90328010 2002-05-11 01:25:00 WD1844−223 REJ, EUVEJ P20405010 2001-04-28 20:01:00 WD1845+683 KUV Z90329010 2002-05-08 20:40:00 Z99003010 2002-09-13 10:39:00 WD1917+509 REJ, EUVEJ Z90330010 2002-05-07 01:36:00 WD1921−566 REJ, EUVEJ A05411110 2000-05-24 22:59:00

(8)

Table 2

(Continued)

WD No. Alt. Name/Cat Observation Start Time (yyyy-mm-dd)

WD1950−432 HS Z90332010 2002-10-04 09:18:00 WD2000−561 REJ, EUVEJ Z90333010 2002-04-15 21:30:00 WD2004−605 REJ, EUVEJ P20422010 2001-05-21 21:09:00 WD2011+398 REJ, EUVEJ M10531020 2003-10-23 12:19:00 WD2014−575 LS210-114 Z90334010 2002-04-16 15:55:00 WD2020−425 REJ, EUVEJ Z90335010 2002-06-07 09:16:00 WD2043−635 BPM13537 Z90337010 2002-04-13 09:10:00 WD2111+498 GD394 M10532010 2002-10-27 16:44:00 WD2116+736 KUV Z90338020 2002-10-28 01:26:00 WD2124+191 HD204188 A05409090 2001-07-12 22:23:00 WD2124−224 REJ, EUVEJ P20406010 2001-10-08 20:26:00 WD2146−433 REJ, EUVEJ Z90339010 2002-10-06 15:14:00 WD2152−548 REJ, EUVEJ M10515010 2002-09-24 03:07:00 WD2211−495 REJ, EUVEJ M10303160 2003-05-24 23:28:00 WD2257−073 HD217411 A05410100 2000-06-28 11:05:00 WD2309+105 GD246 P20424010 2001-07-14 01:26:00 WD2321−549 REJ, EUVEJ Z90342010 2002-07-22 05:10:00 WD2331−475 REJ, EUVEJ M10517010 2002-09-23 08:46:00 WD2350−706 HD223816 A05408090 2001-05-23 16:21:00 ∗G191-B2B obs. M1010201000 1999-10-13 01:25:00 M1030501000 1999-11-12 07:35:00 M1030502000 1999-11-20 07:22:00 M1030401000 1999-11-20 09:02:00 M1030603000 1999-11-20 10:43:00 M1030503000 1999-11-21 06:43:00 M1030504000 1999-11-21 10:03:00 M1030602000 1999-11-21 11:39:00 S3070101000 2000-01-14 09:40:00 M1010202000 2000-02-17 06:10:00 M1030604000 2001-01-09 09:02:00 M1030506000 2001-01-09 09:26:00 M1030605000 2001-01-10 13:20:00 M1030507000 2001-01-10 13:45:00 M1030403000 2001-01-10 15:08:00 M1030606000 2001-01-23 06:08:00 M1030508000 2001-01-23 07:55:00 M1030404000 2001-01-23 11:18:00 M1030607000 2001-01-25 04:46:00 M1030509000 2001-01-25 06:33:00 M1030405000 2001-01-25 09:53:00 M1030608000 2001-09-28 13:50:00 M1030510000 2001-09-28 15:35:00 M1030406000 2001-09-28 17:15:00 M1030609000 2001-11-21 09:54:00 M1030511000 2001-11-21 11:39:00 M1030407000 2001-11-21 13:19:00 M1030610000 2002-02-17 07:27:00 M1030512000 2002-02-17 12:34:00 M1030408000 2002-02-17 17:43:00 M1030611000 2002-02-23 02:05:00 M1030513000 2002-02-23 06:43:00 M1030409000 2002-02-23 08:23:00 M1030612000 2002-02-25 02:17:00 M1030514000 2002-02-25 06:59:00 M1030613000 2002-12-03 21:00:00 M1030614000 2002-12-06 02:30:00 M1030515000 2002-12-06 05:16:00 M1052001000 2002-12-07 21:46:00 M1030615000 2002-12-08 22:36:00 M1030516000 2002-12-09 00:29:00 M1030412000 2002-12-09 03:53:00 M1030616000 2003-02-05 19:14:00 M1030517000 2003-02-05 21:07:00 M1030413000 2003-02-06 00:36:00 S4057604000 2003-11-21 05:21:00 M1030617000 2003-11-23 20:16:00 M1030519000 2004-01-25 21:31:00 M1030415000 2004-01-26 00:51:00

(9)

Figure 1. FUSEspectrum of the hot DA white dwarf, WD 0131−164 (GD 984), with the main photospheric and interstellar features labeled. The wavelengths of all these absorption lines can be obtained from Table2. We note that the line identified as photospheric Si iv 1066.62 Å in this plot may often be contaminated by interstellar Ar i, which occurs at the same wavelength.

(10)

Figure 2. White dwarf heliospheric velocity (vphot) compared to the ISM

velocity (vISM) for the measurements listed in Table3.

calculated velocity differences. Figure 2 compares the white dwarf photospheric heliocentric velocity with that measured for the ISM. There is no particular strong trend, but on average the photospheric velocities are offset by∼15–20 km s−1, a value significantly lower than the 35.2± 20.4 km s−1 obtained by Savage & Lehner (2006). This is illustrated in Figure3, showing the distribution of velocity differences (vPHOT− vISM), which average 20.3± 20.1 km s−1. Within the measurement uncer-tainties, this is close to the 20 km s−1gravitational redshift of a typical 0.54 solar mass (R= 0.017 R) hot white dwarf star.

To monitor consistency, we have examined whether or not the velocities of individual lines vary systematically with wave-length and considered the scatter of the individual data points about the mean. Where we have reliable measurements of lines near O vi we have also compared these velocities with the com-puted means as a further consistency check. For all the stars in our study any discrepancies are well within the∼15 km s−1 limit of the R∼ 20,000 FUSE resolving power.

Measurements of interstellar opacity made with a higher spec-tral resolution than is available with FUSE reveal that the inter-stellar lines of sight are often complex, with multiple compo-nents at different velocities (e.g., Sahu et al.1999). Therefore, the interstellar lines observed in this work might well be blends of more than one component and the observed interstellar veloc-ity the mean of these, weighted by the individual line strengths. The presence of unresolved blends may lead to increased scat-ter in the velocities we measure for individual ISM lines, as the weightings vary from line to line according to differing inter-stellar cloud conditions. However, in this work, we are mostly interested in defining the separation between the ISM and pho-tospheric velocities. The only real issue is that we need to take into account these systematic uncertainties in the analysis.

Table 3 lists the photospheric (where measureable) and interstellar velocities for each star in the sample, determined using the approach outlined above. We do not give statistical errors for these measurements because they are usually very

Figure 3.Number distribution of velocity differences (vPHOT− vISM). The

distribution has a mean of 20.3 km s−1.

small compared to the systematic effects. The largest systematic issue is associated with the absolute velocity scale. However, any offsets apply equally to all our measurements and, as we are only interested in measuring the differences between the O vi, photospheric and ISM velocities, we can safely ignore this. In principle, the instrument resolving power gives an absolute limit (15 km s−1, corresponding to the nominal LWRS resolution of 20,000) to the determination of velocities, but making the assumption that the observed lines should be symmetrical we can usually achieve a factor of∼3 better using a centroid determination algorithm rather than merely measuring the velocity of the minimum flux pixel in the line. Therefore, we assign a general uncertainty of 5 km s−1to all the present velocity measurements.

3.4. Detection and Measurement of Ovi

We have inspected every FUSE spectrum in our sample to look for evidence of the presence of O vi, taking into account the possible uncertainties in its location, depending on whether or not the two possible lines at 1031.912 Å and 1037.613 Å might be located at the interstellar or photospheric velocity (see Tables 3 and 4). We also widened the search to ±0.2 Å (60 km s−1) either side of the range to ensure we did not miss any O vi that might be associated with another velocity component. O vi 1031.912 Å is detected in 41 objects in our sample,≈50% of the objects studied, and the weaker 1037.613 Å line is seen in 16 objects. The FUSE spectra in the vicinity of the O vi line detections are shown in Figure4for the 1031.912 Å line.

The 1037.613 Å line is detected only when 1031.912 Å is present, which is expected from the relative oscillator strengths. The issue of the significance of a detection of either O vi line is an important one, which obviously depends on the intrinsic line strength and the S/N of the data. In many white dwarfs, the O vi lines are strong, and, therefore, obvious in the data, with their

(11)

Table 3

Measured Photospheric and Interstellar Velocities (km s−1)for each White Dwarf where O VI 1031.912 Å is Detected

WD No. vphot vism δv vO vi EW σ EW/σ |vO vi−vphot| |vO vi−vism| O vi? S&L

WD0001+433 7.0 −10.7 17.7 < 23.4 ND WD0004+330 −1.8 −11.0 6.1 1.2 5.0 11.0 9.2 I I WD0027−636 21.8 −5.5 27.3 0.5 28.6 4.6 6.2 21.3 6.0 I+P P WD0050−332 18.4 −14.9 33.4 13.7 6.3 1.7 3.8 4.8 28.6 P P WD0106−358 −12.1 −5.0 −7.1 < 15.3 ND WD0131−164 10.9 −8.6 19.5 19.5 43.2 3.0 14.5 8.6 28.1 P WD0147+674 −15.5 < 26.1 ND ND WD0226−615 25.8 −5.5 31.2 9.3 16.4 4.5 3.6 16.5 14.8 P WD0229−481 23.0 0.9 22.1 < 12.0 ND WD0232+035 4.0 −7.6 11.5 −3.5 8.4 1.4 5.9 7.5 4.1 P WD0235−125 13.7 < 20.6 ND WD0236+498 −4.5 < 19.8 ND WD0252−055 10.4 12.2 −1.8 16.2 23.9 6.7 3.6 5.8 4.0 P WD0320−539 2.5 < 15.8 ND WD0325−857 −10.0 < 5.3 ND WD0346−011 15.8 < 5.7 ND WD0353+284 9.0 25.6 69.9 11.3 6.2 16.6 IR WD0354−368 8.7 < 44.9 ND WD0455−282 57.7 −14.2/−65.5 71.9 59.3 11.3 2.3 5.0 1.6 73.4 IB+P I+P WD0501+524 21.5 14.3 7.2 20.4 4.8 0.5 10.7 1.1 6.1 P P WD0512+326 20.5 −3.9 24.4 25.0 14.4 4.4 3.3 4.4 28.8 P WD0549+158 8.3 1.4 6.8 < 4.0 ND ND WD0603−483 17.5/−34.6 < 16.5 ND I∗ WD0621−376 31.3 13.2 18.1 34.3 4.7 1.0 4.5 3.0 21.1 P WD0659+130 −0.3 < 26.3 ND WD0715−704 −4.8 −11.3 11.6 3.1 3.7 6.5 I I WD0802+413 6.9 13.9/61.4 −6.9 < 33.7 ND ND WD0830−535 11.1 −1.9 31.0 9.0 3.4 13.0 I I WD0937+505 −6.5 40.7 10.7 3.8 I I WD1019−141 11.5 −6.8 18.3 < 10.9 ND I∗ WD1021+266 12.0 −5.3 17.2 13.4 36.4 5.6 6.5 1.4 18.6 P WD1024+326 −14.0 < 51.8 ND WD1029+537 13.8 −22.8 36.6 25.7 63.6 4.4 14.6 11.9 48.5 P WD1040+492 −13.9 −12.6/−47.9 −1.3 45.0 30.3 3.7 8.2 58.9 57.6 IR WD1041+580 −10.4 < 14.6 ND ND WD1057+719 −16.1 19.8 26.3 2.9 9.1 35.9 IR I WD1109−225 −4.9 −8.0 13.8 4.0 3.5 3.1 I WD1234+481 −9.6 −6.4 10.4 3.0 3.5 3.2 I I WD1254+223 −11.8 17.4 15.1 5.0 3.0 29.2 IR I WD1302+597 −6.7 < 17.4 ND WD1314+293 −10.1 19.8 10.5 1.8 5.8 29.8 IR I WD1337+701 −17.2 −12.1 −5.2 < 113.4 ND ND WD1342+442 −13.1 −25.6 12.5 −12.2 72.3 8.0 9.1 0.9 13.3 P WD1440+753 −14.0 < 11.9 ND ND WD1528+487 −27.3 −34.3 23.8 4.1 5.8 7.0 I I WD1550+130 −9.9/43.8 < 23.0 ND WD1603+432 −34.3 < 16.1 ND I WD1611−084 −35.3 −21.0/40.6 −14.3 −24.2 53.1 4.0 13.3 11.1 3.2 P WD1615−154 −68.3 < 12.0 ND ND WD1620+647 −31.7 < 17.2 ND ND WD1631+781 −11.8 −8.0 6.9 2.1 3.3 3.8 I I WD1635+529 −22.5 < 16.3 ND WD1636+351 −31.1/24.9 < 15.1 ND I∗ WD1648+407 −28.4 −18.0 21.9 9.3 2.4 10.4 I I WD1725+586 −37.8/25.4 18.7 53.1 5.2 10.2 56.5 IR WD1800+685 −29.0 −18.9 11.7 2.1 5.5 I I WD1819+580 −8.4 −24.8 16.4 < 13.7 ND WD1844−223 −35.5 < 7.4 ND ND WD1845+683 −28.2 < 32.2 ND ND WD1917+509 3.1 −29.3 32.4 < 12.5 ND ND WD1921−566 −30.1 < 22.2 ND WD1942+499 7.5 −27.8 35.3 6.4 15.7 3.1 5.1 1.1 34.2 P P WD1950−432 −26.0 −8.4 −17.6 0.3 17.5 5.0 3.5 26.3 8.7 I I WD2000−561 −9.6 −26.2 16.7 −15.3 25.4 6.9 3.7 5.7 11.0 P P WD2004−605 −27.2 −18.0 9.4 2.8 3.4 9.2 I I WD2011+398 11.1 −12.3 23.4 13.9 62.6 3.5 18.0 2.7 26.1 P

(12)

Table 3

(Continued)

WD No. vphot vism δv vO vi EW σ EW/σ ϕvO vi−vphotϕ ϕvO vi−vismϕ O vi? S&L

WD2014−575 −29.8 < 29.0 ND ND WD2020−425 −25.4 < 42.0 ND WD2111+498 24.7 −20.0 44.7 13.1 20.1 4.2 4.8 11.7 33.1 P P WD2116+736 −17.6 −3.5 9.9 2.6 3.8 14.1 I I WD2124+191 −19.7 < 44.7 ND WD2124−224 29.7 −28.4 58.1 −16.4 20.1 4.7 4.3 46.1 12.0 I I WD2146−433 27.9 −13.1 41.0 10.4 20.3 4.5 4.6 17.5 23.5 P P WD2152−548 −3.9 −2.6 90.0 6.6 13.7 1.3 I WD2211−495 23.9 −8.9 32.8 18.0 15.4 1.2 13.4 5.9 26.9 P P WD2257−073 −10.2 30.8 19.1 6.2 3.1 41.0 IR WD2309+105 −33.1 −28.7 −4.4 < 5.1 ND ND WD2321−549 17.5 −16.6 34.2 −1.6 44.5 6.3 7.0 19.1 15.0 I+P P WD2331−475 28.3 −3.0 31.2 3.0 13.2 2.4 5.6 25.2 6.0 I+P P WD2350−706 40.7 −6.4 47.1 10.6 45.1 3.6 12.5 30.0 17.0 I+P

Notes.Tabulated above are the difference between photospheric and ISM velocities; the wavelength, velocity, equivalent width (mÅ), and significance of the O vi 1031.912 Å (EW/uncertainty in EW); absolute differences between the O vi and photospheric/ISM velocities; indication of the nature of the O vi absorption (P: photospheric, I: interstellar at vism, IR: interstellar redshifted w.r.t. vISM, IB: interstellar blueshifted w.r.t. vISM, ND: non-detection); nature of

the O vi as determined by Savage & Lehner (2006). The data are grouped by the nature of the O vi detected in these stars. Symbol “∗” for these stars denotes the reason for the discrepancy between this work and Savage & Lehner is their adoption of a 2σ detection threshold.

Figure 4.Normalized profiles vs. the heliocentric velocity of the local ISM C ii and O i absorption lines (lower panel) compared to the O vi region. The vertical dashed line marks the photospheric velocity and the dash-dotted line the mean interstellar velocity for each star. We mark the statistical error bars in the top panel of each plot to illustrate the significance of the O vi detections we report but have left them out of the bottom panel for clarity.

location well determined. However, weaker lines, particularly those close to the noise limit, are not so readily identified and separating real O vi detections from “noise” features becomes

problematic. We have measured the velocities of all detected O vi lines as described in Section3.3. We also determined their equivalent widths (EWs) in the standard way, as the area of

(13)

Figure 4.(Continued)

Table 4

Stars Where O vi 1037.613 Å is Detected

WD No. Star vO vi(1031.912) O vi? vO vi(1037.613) EW σ EW/σ

WD0027−636 REJ, EUVEJ 0.49 I+P 16.79 28.07 4.53 6.20

WD0131−164 GD984 19.46 P 18.59 50.29 2.47 20.36

WD0252−055 HD18131 16.24 P 27.26 28.47 4.88 5.83

WD0455−282 REJ, EUVEJ 59.27 IB+P −65.35 11.40 1.91 5.97

WD0512+326 HD33959C 24.96 P 12.20 14.24 2.82 5.05 WD0621−376 REJ, EUVEJ 34.34 P 25.15 2.90 0.85 3.41 WD1021+266 HD90052 13.36 P 12.27 32.08 4.74 6.77 WD1029+537 REJ, EUVEJ 25.68 P 24.09 50.47 3.46 14.59 WD1342+442 PG −12.23 P −20.14 69.07 6.96 9.92 WD1528+487 REJ, EUVEJ −34.34 I −32.52 18.78 4.00 4.70 WD1611−084 REJ, EUVEJ −24.20 P −26.24 44.96 3.93 11.44 WD1615−154 G153−41 46.72 ND 20.17 7.67 3.00 2.56 WD1725+586 LB335 18.70 I 18.09 45.75 4.94 9.26 WD2011+398 REJ, EUVEJ 13.89 P 11.18 44.76 2.78 16.10 WD2124−224 REJ, EUVEJ −16.39 I 11.25 5.04 2.93 1.72 WD2152−548 REJ, EUVEJ −2.61 I −8.99 55.42 4.78 11.59 WD2211−495 REJ, EUVEJ 18.04 P 18.20 13.71 1.16 11.82

the line falling below the nominal continuum flux level. The associated uncertainty in the EW is estimated from the size of adjacent noise features in the spectrum, which allows us to assign a measure of significance for each “detection”. Where there is no clear detection, we have determined a 3σ upper limit to the line strength from the typical size of noise in the vicinity of the O vi location.

Table3includes all O vi 1031.912 Å measurements, giving the measured wavelength, calculated radial velocity of the line, the EW, its uncertainty, and its significance (the EW divided by the uncertainty). However, where only an upper limit to the O vi line strength is obtained, we only give that EW as the other parameters are not relevant. Of the “detected” lines, the measured significance ranges from a maximum∼18σ down to

(14)

Figure 4.(Continued)

our 3σ threshold. Any detection below 3σ is not considered to be significant and is excluded from further analysis and discussion, apart from as an upper limit to the amount of O vi present. Similar data for the O vi 1037.613 Å detections are given in Table4. All the upper limits for detection of O vi 1031.912 Å are listed in Table3but we provide no upper limits for the non-detections of the 1037.613 Å line because these far outnumber the reliable detections and, since they are calculated from the general noise level in the spectrum, will be the same as the values listed for the 1031.912 Å line.

A number of stars in the sample show evidence for multiple stellar components. Where these are clearly detected we note the velocity of the additional component in Table3. There is no clear evidence of any secondary O vi component associated with any of these weaker features. However, in WD0445−282, the inter-stellar O vi absorption is broad and spans the weaker feature. There is also a hint in Figure4that there might be a secondary O vi feature in WD1725+586, but it does not formally exceed our applied 3σ threshold. We discuss the particular properties of some individual stars and lines of sight in theAppendix.

4. DISCUSSION

4.1. Determination of the Origin of the Ovi

As described earlier in this paper, there is strong evidence from other observations that relatively high-ionization stages are seen in white dwarf photospheres at temperatures below

those limits expected from stellar atmosphere calculations, due to stratification of material in the stellar envelopes. Therefore, when any metals are present in the atmosphere of a white dwarf we are unable to rely on the ionization stage and stellar temperature as a means of discriminating between a photospheric and interstellar origin for the O vi. The only completely reliable method we can then use is to compare the O vi radial velocity with the values obtained for the photosphere and ISM in each star.

How well we can discriminate between a photospheric and interstellar origin depends on the separation of the two velocity components and the uncertainty in the velocity measurements. In Section3.3, we estimated that the systematic and limiting un-certainty in the velocity measurements was∼5 km s−1. Hence, the photospheric and interstellar components must be at least separated by this amount for us to be able to discriminate. How-ever, this is a somewhat na¨ıve and likely underestimate of the magnitude of velocity difference that can be excluded with high confidence. First, we are looking at velocity differences and, therefore, the combined uncertainty of two velocity measure-ments needs to be taken into account. Furthermore, the veloc-ity measurement error adopted corresponds to a ∼1σ value. Therefore, there is a 68% probability that the true velocity lies within the quoted range of uncertainty. If we are to claim that a particular O vi line resides at either photospheric or inter-stellar velocity, we need to be able to rule out an association with the other component at high confidence. Adopting a 2σ

(15)

Figure 4.(Continued)

(10 km s−1) uncertainty yields a 95% confidence limit. For the purposes of deciding the nature of the O vi components we look for a 14 km s−1(√2× 10 km s−1) minimum separation to rule out an association.

4.1.1. Detection of Photospheric O vi

As discussed in Section 1, O vi can be present in white dwarf photospheres and is not necessarily restricted to high-temperature objects. The detections are only designated photo-spheric if the detected line resides close to the measured stellar velocity and is well separated from the location of the expected interstellar component. Good examples are WD0131−164 or WD2211−495. Slightly problematic are objects such as WD1029+537. In this example, the core of the O vi line is 1–2 wavelength bins redward of the photospheric velocity. How-ever, this remains within the overall measurement uncertainties and so this is also classified as a likely photospheric detec-tion. In any case, there is sufficient ambiguity to rule it out as interstellar in the context of this analysis. In many stars, we have a clear detection of O vi but the velocity difference be-tween the ISM and photosphere is insufficiently large to make a clear statement about the nature of the detection. For example, in WD0501+524, the two components are virtually coincident, while in WD1611−084 they are separated by ∼15 km s−1, but still too small, when compared to the width of the line and the accuracy of the velocity measurements to assign an origin.

In the hottest white dwarfs, it is most likely that the O vi is photospheric in origin. Below ∼45,000–50,000 K this is not necessarily the case, but since we know of several examples (WD0050−332 is one) where highly ionized metals are present at lower temperatures we cannot be definitive about the origin of the O vi.

4.1.2. Detection of Interstellar O vi

Any detection recorded at a velocity other than the photo-spheric value we attribute to interstellar material. In 43 out of 80 white dwarfs a thorough search of the FUSE spectrum reveals no detectable photospheric material. If the O vi were photospheric, we would expect to see evidence of some other species. Therefore, we conclude that any O vi present is most likely to be interstellar in these stars. For most detections, the velocity of O vi was coincident, within measurement uncertain-ties, with that of the cooler ISM, as determined from neutral or singly ionized species. However, in several cases the O vi ve-locity was redshifted with respect to the ISM, as illustrated by WD1057+719 and WD1040+492. In one star, WD0455−282, the O VI is blueshifted.

4.1.3. Detection of Both Interstellar and Photospheric O vi

Interestingly, there is also a clearly resolved photospheric component in the spectrum of this WD0455−282. There may also be a weak component at the ISM velocity, but this cannot

(16)

Figure 4.(Continued)

be resolved from the broad blueshifted feature. There are three other objects where both interstellar and photospheric compo-nents appear to be present. However, the separation of the ISM and photospheric velocity components is not large enough to completely resolve these. For example, WD2321−549 exhibits a broad feature which appears to have two minima aligned with each velocity component. If there were no interstellar compo-nent or if it was much weaker than the photospheric contribution, the features would be strongly skewed toward the photospheric velocity. Hence, this is a clear indication that this is a blend of two lines. Three other stars, WD0027−636, WD2331−475, and WD2350−706, show similarly broad lines, which are good evidence for two blended components, but with less distinct separation. For these stars we record the full strength of the line from both contributions in Table 3 but take into account the photospheric contamination, which is approximately 50% of the total in these cases, by fitting an additional Gaussian pro-file in estimating the interstellar column and volume densities in Table5. As a result, there will be some additional system-atic uncertainty in the measurements. We have not included this in the error estimates but indicate the increased uncertainty in Tables5and6.

Taking all the above into account we label the most likely origin of the O vi in Table3as follows: definitely or possibly photospheric (P), interstellar at vISM(I), interstellar redshifted relative to vISM (IR), interstellar blueshifted relative to vISM (IB), not detected (ND). The interpretation of Savage & Lehner

(2006) is also shown for comparison with our results. It is im-portant to note that when we used the designation “P,” we are indicating that the star contains other photospheric components which make it likely that there is a significant photospheric contribution to the O vi feature. However, this does not defini-tively exclude the possibility of an interstellar component being present, hidden by or blended with a photospheric contribution when the velocity separation is below the instrument resolution. In four stars, (WD0027−636, WD2321−549, WD2331−475, and WD2350−706), we see broad O vi components which we interpret as blended photospheric and interstellar lines and des-ignate “I+P,” even though not fully resolved. For these stars, the photospheric and interstellar components are severely blended and for our estimate of the interstellar contribution, we make the assumption that the photospheric and interstellar compo-nents are roughly equal in strength. The result is of course very uncertain and that is why we do not estimate any errors and use the symbol “∼” to indicate the high level of uncertainty in Tables5and6. In other examples, where the photospheric fea-ture is dominant the presence of any ISM component is much more ambiguous and we are not able to carry out the same analysis.

4.2. Comparison with Other Studies

As discussed extensively already in this paper, there have been two previous studies of interstellar O vi using white dwarfs as background sources, by Oegerle et al. (2005) and Savage &

(17)

Figure 4.(Continued)

Lehner (2006). Essentially, the Oegerle et al. sample is a subset of that of Savage & Lehner. Therefore, in this section we will just compare our results with the more recent paper. There is considerable agreement between the analysis we have conducted here and that of Savage & Lehner. We have 26 detections of interstellar material, compared to their 21 for the stars that are in common with this study. Since we apply a higher (3σ , cf. 2σ ) detection threshold a few detections reported by Savage & Lehner (WD0603−483, WD1019−141, WD1603+432, and WD1636+351) are not included in our list, although we do obtain similar EW and significance in each case. There are two stars in the Savage and Lehner sample (WD0027−636 and WD2321−549) where they designated the O vi as photospheric but which appear to have a blended ISM component (also noted by them) and where we have estimated the strength of the ISM component.

Despite the fact that the sample of DA white dwarfs studied in the work is almost a factor of 2 larger than that of Savage & Lehner, we have detected interstellar O vi in only eight additional lines of sight a∼25% increase when counting just the 3σ and above detections, indicating a rather lower “success rate” in detecting interstellar O vi. Three of these (WD0027−636, WD2331−475, and WD2350−706) are also apparent blends of ISM and photospheric contributions. Hence, this may be partly due to Savage & Lehner’s a priori exclusion from their analyses previously known stars where the O vi is thought to be contaminated or where the S/N is not adequate, which we did consider here.

4.3. OviColumn and Volume Densities

We have reliable temperature and gravity estimates for all the white dwarfs in our sample, obtained from either Balmer or Lyman line analyses, as listed in Table 1. Coupling this information with visual magnitude and/or flux data allows us to estimate a distance to each object. This process depends on the assumption that the standard mass–radius relation for a typical DA white dwarf is correct (see Barstow et al.2005, for further discussion on this topic) but with that, and the additional proviso that there are no hidden companions, yields distances accurate to a few percent (Holberg et al.2008) and which we have listed in Table1. Consequently, we can examine both the distribution of column and volume densities for the O vi detections and upper limits. The upper limits obtained are listed in Table3and span a range of EWs from 4.0 mÅ for the very highest S/N spectra to 113 mÅ, where the data have poor S/N. Since all the detected O vi lines are relatively weak, it is reasonable to assume that they are not saturated. Therefore, column densities have been calculated using the standard linear curve-of-growth method for unsaturated lines:

N = Wλ/(8.85× 10−21f λ2),

where N is the column density (cm−2), Wλ is the EW of the observed line/upper limit (mÅ), and f is the oscillator strength of the line (0.133 for O vi at λ= 1031.93 Å). The average volume density along the line of sight (n, cm−3) is then calculated simply by dividing by the distance to the star. The results of these

(18)

Figure 4.(Continued)

calculations are summarized in Table5and in Figures5and6, which show column and volume density as functions of stellar distance, respectively. The black diamonds and error bars are the measured values while the gray diamonds are the upper limits. For a number of stars with the highest S/N the implied densities determined from the upper limits have lower values than the actual detections. Hence, these limits provide tight constraints on the amount of hot gas present along those particular lines of sight.

One possible model for the LB is that it is full of hot X-ray emitting gas. If this material were uniformly distributed, we would expect to see O vi detections exclusively from transition temperature gas at the interface with cooler material, near the bubble boundary. Closer examination of Figure 5 shows that several of the well-constrained upper limits are obtained for stars that are closer to the Sun than the main group of detections. However, at least half the upper limits are at similar distances to stars with interstellar detections. Indeed, one non-detection (WD0229−481) is the most distant object in this group. This is a strong indication that the distribution of material around the Sun is far from uniform and that interfaces are not seen in all directions.

We already know from earlier studies of the LISM, culmi-nating in the work of Lallement et al. (2003), that the boundary of the LB is not equidistant from the Sun in all directions. Nor is it a simple shape. So perhaps it is not surprising that there is significant variance in the measured column densities and upper

limits. For sources within the LB, we might expect the conver-sion to volume density to produce a more coherent picture. In fact, as can be seen in Figure6, this is not so, with very low col-umn densities being recorded over some of the larger distances. Again, the picture is one of extreme patchiness, with volume densities varying by an order of magnitude between stars at similar distances.

4.4. The Spatial Distribution of the Local OviAbsorption

Previous authors have already described the distribution of O vi absorption within 200 pc of the Sun as being “patchy” (Oegerle et al.2005; Savage & Lehner2006). If O vi forms at the evaporative interfaces between cool clouds and the surrounding hot million degree (soft X-ray emitting) LB gas, then the presence of tangled or tangential magnetic fields could quench thermal conduction over a large portion of a cloud’s surface and thus O vi would only be formed in “patches” (Cox & Helenius

2003). However, Welsh & Lallement (2008) have suggested that many sight lines along which local interstellar O vi absorption (d < 100 pc) and O vi emission have been detected seem to preferentially occur at high galactic latitudes, in regions that also possess high levels of SXRB emission. To test this possible spatial correlation claim further with our new data, in Figure7

we show the galactic distribution of the white dwarf sight lines within 120 pc together with B star sight lines (also within 120 pc) listed by Welsh & Lallement (2008) and Bowen et al. (2008) toward which O vi absorption has been detected with

Figure

Figure 1. FUSE spectrum of the hot DA white dwarf, WD 0131 − 164 (GD 984), with the main photospheric and interstellar features labeled
Figure 2. White dwarf heliospheric velocity (v phot ) compared to the ISM velocity (v ISM ) for the measurements listed in Table 3.
Figure 4. Normalized profiles vs. the heliocentric velocity of the local ISM C ii and O i absorption lines (lower panel) compared to the O vi region
Figure 4. (Continued) Table 4
+4

Références

Documents relatifs

Statistical results on the current distribution and the maximum current as a function of the density of defects in the network are compared with theoretical

The proposed attribution of the diffuse interstellar bands at 9577 and 9632 Å to electronic transitions of the buckminsterfullerene cation (i.e. C + 60 ) was recently supported by

Retrouve les chiffres qui ont fondu à la chaleur du

Rails utilise sqlite3 comme base de données par défaut, mais vous pouvez générer une nouvelle application rails avec une base de données de votre choix. Ajoutez simplement l'option

[r]

Mass ejections and morphological changes observed in the par- ticular Khonsu bank paint the variety of mechanisms in broad strokes that trigger outbursts on the surface of

The One Lakh Housing Scheme (OLHS), carried out by the Kerala state government between 1972 and 1976 has been noted by many central and state level officials,

Besides the eight imino signals previously detected in Ab/A, additional imino protons possibly associated with T19 in Ab/G and Ab/C (red arrows, Supplementary Figure S7A and B) and